janhoy commented on code in PR #5112: URL: https://github.com/apache/solr/pull/5112#discussion_r4335515592
########## solr/core/src/java/org/apache/solr/servlet/NodeTargetProxyFilter.java: ########## @@ -0,0 +1,138 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.solr.servlet; + +import jakarta.servlet.FilterChain; +import jakarta.servlet.ServletException; +import jakarta.servlet.UnavailableException; +import jakarta.servlet.http.HttpServletRequest; +import jakarta.servlet.http.HttpServletResponse; +import java.io.IOException; +import org.apache.solr.client.solrj.jetty.HttpJettySolrClient; +import org.apache.solr.common.SolrException; +import org.apache.solr.common.cloud.ZkStateReader; +import org.apache.solr.core.CoreContainer; +import org.eclipse.jetty.client.HttpClient; + +/** + * If the request carries the {@link #TARGET_NODE_HEADER} header naming a different live node, + * proxies the request there verbatim and returns; otherwise passes the request through. Unlike + * {@link HttpSolrCall}'s collection/core-driven proxying, this works for any endpoint, including + * node-level ones (e.g. {@code /solr/admin/**}, {@code /api/node/**}), so a request can be pinned + * to a specific node even when reached through a load balancer that hides which node answers. + */ +public class NodeTargetProxyFilter extends CoreContainerAwareHttpFilter { + + /** + * Header naming a live node (as it appears in ZK live_nodes) to proxy this request to. Two + * shorthand forms are also accepted, both resolved relative to the current node: a value of all + * digits replaces the trailing {@code -<digits>} ordinal in the current node's hostname (e.g. + * {@code 8} turns {@code myhost-3.mydomain.com:3000} into {@code myhost-8.mydomain.com:3000}); + * otherwise, a value with no '.' is treated as a bare peer hostname, with the current node's + * domain suffix and port appended. + */ + public static final String TARGET_NODE_HEADER = "X-Solr-targetNode"; + + @Override + protected void doFilter(HttpServletRequest req, HttpServletResponse resp, FilterChain chain) + throws IOException, ServletException { + // also accept it as a query param on GET requests, for convenience when a header can't + // easily be set (e.g. a browser address bar) + String targetNode = req.getHeader(TARGET_NODE_HEADER); + String queryString = req.getQueryString(); + if (targetNode == null + && "GET".equals(req.getMethod()) + && queryString != null + && queryString.contains(TARGET_NODE_HEADER)) { + targetNode = SolrRequestParsers.parseQueryString(queryString).get(TARGET_NODE_HEADER); + } + if (targetNode == null || targetNode.isBlank() || !proxyToNode(targetNode, req, resp)) { + chain.doFilter(req, resp); + } + } + + /** Returns true if the request was proxied to {@code targetNode}. */ + private boolean proxyToNode(String targetNode, HttpServletRequest req, HttpServletResponse resp) { + CoreContainer cores; + try { + cores = getCores(); + } catch (UnavailableException e) { + throw new SolrException(SolrException.ErrorCode.SERVER_ERROR, "Core Container Unavailable"); + } + if (!cores.isZooKeeperAware()) { + throw new SolrException( + SolrException.ErrorCode.BAD_REQUEST, TARGET_NODE_HEADER + " header requires SolrCloud"); + } + String ownNodeName = cores.getZkController().getNodeName(); + String resolvedNode = resolvePeerShorthand(targetNode, ownNodeName); + ZkStateReader zkStateReader = cores.getZkController().getZkStateReader(); + if (!zkStateReader.getClusterState().getLiveNodes().contains(resolvedNode)) { + throw new SolrException( + SolrException.ErrorCode.NOT_FOUND, "Unknown or non-live node: " + resolvedNode); Review Comment: It could be that the caller is unlucky and hits a glitch where a node is restarting. Then it is technically not the caller's fault. But it would require us to consult the wider cluster state to know if the node COULD EVER be a live node. A 400 would not be retryable since it's an error by the caller to try to reference a non-approved node. But if not live right now, but in 5s it would be retryable. Problem I believe is that unless there's a collection/replica registered on nodeX, we'd have no way to know that the node was part of the cluster 30s ago and will come bac in 5s. Let's keep it simple and do 400. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
